So what is the core difference between agent first search versus like consumer or like a human first search?

That's a great question.

How long do we have? But it's really, really great question.

Let me talk about It's really interesting.

We have an amazing data science team that is AI eval team that is constantly looking at how do we improve our products, right? And about nine months ago, they had this really interesting observation that There were two very different types of search patterns occurring on our system.

Right.

One was what we call natural language.

How is the enterprise typically use u.com and like what would be like the end to end solution kind of like look like?

Is the issue mostly because of the model capability or the data access, or do you think governance or user experience or what are, or is this something completely different? Because here in your conversations, what do you see the biggest challenges these enterprises are facing when they're using this AI assistance or agents in real business workflows?

Yeah, it's an interesting question.

There's many different things going on.

Every enterprise is a little bit different.

Let me talk through some themes that I see, and sometimes they're contradictory themes.

So let's talk about what's very trendy in the news these days.

I'm spending too much on AI, right? You've heard that from Uber, CTO, those type of things.

Are we over-focusing more on the model and under-focusing on the product design, workflows, evaluation, or overall user trust itself?

for your world can i ask are you seeing a lot of companies coming into you.com where they've got individuals in the organization organization are excited want to be building agents want to be building workflows sort of hacking and slashing away privately on their devices to try and make this happen and they're looking to now step up and build agents purposefully deliberately and securely at a corporation level and what does agent management look like as we're all going on this new journey of managing agents typically for the first time in our careers

yeah that's a great question i think what you know you're hitting upon a lot of interesting points right i think that to your point Six months ago, the story was go off and experiment, go off and see how you can improve our core business metrics using AI, right? And you saw that happening everywhere.

Now the story very much, you know, there's the famous statements from some of the folks at Uber saying, we spent our entire token budget for the year in three months or four months or whatever it might be.

And what they're starting to recognize is that there's like two conflicting goals.

Goal number one is explore the capability overhang, explore what AI can bring.

But then there is conflicting goal number two, which is doing that in a 10,000 person organization leads to real costs that must be justified.

And so what you're finding is folks are starting to explore this tension to say, perhaps, you know, there's a subset of the world that is now focused on exploring that capability overhang.

Staying on this point, Saurabh, when you're looking to recruit product talent in to you.com, what is it that you're looking for that's different around either mindset or skill set from product managers today versus, say, three, four years ago?